1 / 51

The Link Layer: Framing, Addressing, and Forwarding

The Link Layer: Framing, Addressing, and Forwarding. GZ01 Networked Systems Kyle Jamieson Lecture 4 Department of Computer Science University College London. The link layer. The link layer (L2) has the job of transferring a datagram point-to-point over a link.

lluvia
Télécharger la présentation

The Link Layer: Framing, Addressing, and Forwarding

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. The Link Layer: Framing, Addressing, and Forwarding GZ01 Networked Systems Kyle Jamieson Lecture 4 Department of Computer Science University College London

  2. The link layer The link layer (L2) has the job of transferring a datagram point-to-point over a link Unit of data is called a frame at the link layer

  3. The link layer: Context • Datagram transferred by different link protocols over different links on its path • Each link protocol provides different services • Reliability • Error correction • Half/full duplex

  4. Today • L2 addressing and interaction with L3 • Link layer switching • Learning switches • VLANs • Spanning Tree Protocol • The Point-to-Point Protocol (PPP) • Link virtualization: MPLS

  5. Comparing addressing schemes • Network layer address (IP address) • Function: move datagram to destination network • 32-bit address, dotted quad notation a.b.c.d where each component is an eight-bit unsigned integer • Hierarchical address space • Link layer address (MAC address, Ethernet address): • Function: move frame from one point to another point on the same network • 48-bit address (in most LANs) • Burned in NIC ROM, also sometimes software settable • Usually a flat address space

  6. Link-layer addressing • Each adapter on LAN has unique link layer address, my_station_id • Special broadcast address broadcast_id • Each L3 protocol registers itself in net_handler array • L2 address-related functionality for inbound frames: • Test received_frame.destination against my_station_id or broadcast_id, drop if neither match • handler net_handler • [received_frame.net_protocol] • call handler(received_frame.payload)

  7. Link layer addresses aren’t routable • Network layer routes from one network to another (link layer address)

  8. Mapping network to link addresses • Goal: Translate from nework_send(data, length, IP, N) to link_send(data, length, IP, enet/18) • Name-mapping table: example of soft state

  9. The Address Resolution Protocol (ARP) L sending a datagram to N: L: network_send(data, length, IP, N) L: link_send(“where is N?”, length, ARP, broadcast_id) N: link_send(“N is at enet/18”, length, ARP, enet/17) L: link_send(data, length, IP, enet/18) ARP table at L: Internet Address enet/ station TTL N 18 20 min

  10. ARP’s role in L3 routing • Walkthrough: send IP data from L to E via router K • Router K • Two ARP tables: one for each subnet • Two interfaces (interface ids 6, 4): one for each subnet, with two link-layer addresses (19, 27)

  11. Example: routing between subnets L sending a datagram to E: L: network_send(data, length, IP, E) L: link_send(data, length, IP, enet/28) L: ARPs for K’s link layer address (enet/19) L: link_send(data, length, IP, enet/19) K: ARPs for E’s link layer address (enet/28) on link/4 K: link_send(link/4, data, length, IP, enet/28)

  12. Hubs • Replicating device: bits coming in go out all ports • Operates at the physical layer (L1) • Doesn’t run CSMA/CD: immediately replicates bits • Collisions possible between all hosts • No frame buffering at the hub • Physical limitations: can’t grow collision domain too big Hub

  13. Ethernet nowadays: Switched A • Hosts have dedicated, direct connection to switch • Switches buffer frames • Ethernet protocol used on each incoming link, but no collisions • Each link in own collision domain • Switching: A → A’ and B → B’ simultaneously, without collisions F B 1 2 6 3 4 5 Switch S C E D

  14. The switch table • How to tell which port to forward packet? • The switch table: LAN addr Port TTL • Switch table initially empty • Example of “soft state” • How to build up switch table entries? A F B 21 22 26 1 2 6 3 Switch S 4 5 23 25 24 C E D

  15. Building the switch table: Self-learning • Switch table: (LAN addr, port id, TTL) • Receive a packet from LAN addr enet/src to LAN addr enet/dst: • Add enet/src to switch table with incident port id • If enet/dst is broadcast_id: forward out all ports except id • If enet/dst in switch table, lookup entry and forward to resulting port id • Otherwise, forward out all ports except id • Periodically flush link table entries based on TTL

  16. Building up the switch table: example • A S: [A (enet/21)  D (enet/24)] S  (all ports) • B  S: [B (enet/22)  C (enet/23)] S  (all ports) • D  S: [D (enet/24)  B (enet/22)] S  port 2 only Switch table: LAN addr Port TTL enet/21 1 20 min enet/22 2 20 min enet/24 4 20 min A F B 21 22 26 1 2 6 3 Switch S 4 5 23 25 24 C E D

  17. Hubs versus switches Hub Switch Link-layer (L2) device Selectively forward frame to one or more outgoing links CSMA/CD MAC Buffers frames Hosts are unaware of presence of switches No configuration needed • Physical layer (L1) device • Replicating device: bits in go out all ports • No medium access control • No frame buffering • Hosts are unaware of presence of hubs • No configuration needed

  18. Interconnecting LANs • Switches can connect LANs as well as hosts • Sometimes called bridges in this context S4 Link layer addresses 1 3 S1 2 S3 1 1 11 1 S2 2 21 2 4 3 A 4 3 12 13 22 D 23 B C F E

  19. Interconnecting LANs: example Suppose C sends frame to F, F responds to C S4 Link layer addresses 1 3 S1 2 S3 1 1 11 1 S2 2 21 2 4 3 A 4 3 12 13 22 D 23 B C F E What traffic gets sent, and which additions to switch tables? S1: enet/13: port 4, enet/23: port 1 S4: enet/13: port 1, enet/23: port 2 S2: enet/13: port 1, enet/23: port 4 S3: enet/13: port 1

  20. Interconnecting switches: loops • S1, S4, S2 form loop, packets forwarded around forever! • Why do loops form? • Inadvertently: many people responsible for network, one person adds a bridge • Intentionally: more connections adds redundancy for failure • Consequence: Can’t learn the direction of a source if it’s in more than one direction, so bridge learning algorithm breaks S4 Link layer addresses 1 3 S1 2 S3 1 1 11 1 S2 2 5 2 4 3 A 4 3 12 13 22 23 B C F E

  21. The spanning tree protocol (STP) • Manager at DEC asked Radia Perlman to build a switch (bridge)to connect two Ethernets • Perlman’s idea: Switches agree on a spanning tree • Subset of the topology that is connected • Loop-free • Some path between every pair of LANs • Implementers at DEC resisted (wanted simplest possible design), first customer site connected bridge to one Ethernet twice, generating a broadcast storm • Mechanism: • Switches block some ports from sending or receiving data • Switches continue using the learning switch algorithm to forward over the spanning tree

  22. Spanning tree protocol (STP): Outline • Elect one root switch (switch with the lowest ID) • Compute shortest paths tree from root to each switch S • Note which port of S is on path to root: root port (R) • All switches connected to a LAN choose a designated switch to forward frames to root (switch on path to root from LAN) • e.g. switch 2 is designated for the LAN below • i.e., each port decides if it is a designated port (D) • Block all other ports: blocked port (B) Port number LAN 1(R) 1(D) 2 3 Switch ID Switch

  23. STP: Messages and switch state • All switches exchange configuration messages switch X sends: (Root identifier, distance to root, X) • Configuration messages are never blocked • Switch X generates configuration messages periodically with a “clock tick” timer, initially sending (X, 0, X) • State at each switchX: • Root identifier (initially X) • Configuration message to send (initially (X, 0, X)) • State at each port: • Forwarding data traffic or blocking data traffic (initially forwarding) • “Best” configuration message + age of that message (initially empty)

  24. Electing a per-LAN designated switch • Designated port rule: At a switch, for each port p • Consider all configuration messages received on port p and the configuration message the switch would send • If receive “better” configuration message on a port p, don’t send configuration messages on port p, otherwise p is designated, send on p • Rule for comparing configuration messages: (R1, d1, X1) better than (R2, d2, X2) if R1 < R2or (R1 = R2and d1 < d2) or (R1 = R2and d1 = d2and X1 < X2) • E.g.: 2 sends (2, 0, 2); 3 sends (3, 0, 3); (2, 0, 2) better than (3, 0, 3) • Switch 2, port 1 sends configuration message • Switch 3, port 1 does not send configuration messages 1(D) 1 2 3

  25. STP: initial startup phase 1(D) • Initially, X generates configuration messages periodically, sending (X, 0, X) • All ports designated and forwarding • Result: For each LAN, one attached switch (the designated switch) transmits configuration messages 87 2 78 1(D) 5 1 4 2(D) 90 3 1(D) 2 2 3(D) 66 5 1 2 101 2 1 2 2(D) 1(D)

  26. Calculating root ID and cost to root • Switches continuously take the following steps: • Root ID rule: Root ID r at switch X is the minimum of X and root IDs received at all ports • Calculating distance to root d at switch X: • If X is the root (X = r), d = 0 • Otherwise, X is not the root (X ≠ r): • d = one plus the minimum cost from configuration messages received on all ports (transmitter field breaks ties). Suppose this comes from port p. • Root port rule: Designate port p as a root port • Switch X’s configuration message is now (r, d, X). Reapply designated port rule on all ports • Blocked port rule: Don’t forward data to or from a port if it is not a designated port or a root port

  27. STP: calculating root ID (2,2,78) 1(D) (2,2,87) 87 (2,1,90) 5(D) 2(B) 78 1(R) 1(B) 4(D) (2,1,90) 2(R) 90 (2,1,66) 1(D) 3(B) 2(R) 2(R) 66 5 1(R) 3(D) (2,1,5) 2(B) (2,2,101) (2,0,2) 2(D) 101 1(D) (2,0,2) 2 2(B) 1(R) 90’s root ID r: 2, d = 1 90’s message: (2, 1, 90) 5’s root ID r: 2, d = 1 (port breaks tie) 5’s message: (2, 1, 5)

  28. STP: Handling topology changes • Configuration messages also contain age: (r, d, X, age) • age = 0 when sending configuration message • Best configuration message for each port contains age • Age incremented each unit of time • If age reaches some threshold (max age), discard that configuration message and recalculate using all rules • Recalculate when: • Receive better or newer configuration message on port p: overwrite existing configuration message • Timer ticks: increment message age in stored messages for each port

  29. STP: Handling failures (2,2,78) 1(D) 87 (2,1,90) 5(D) 2(B) 78 1(R) 1(D) (2,1,90) 4(D) (2,1,90) 2(R) 90 3(B) 2(R) 2(R) 66 5 1(R) 3(D) (2,1,5) 2(B) (2,0,2) 2(D) 101 1(D) (2,0,2) 2 2(B) 1(R) 90’s root ID r: 2, d = 1 90’s message: (2, 1, 90)

  30. STP: Handling new bridges (2,2,78) 1(D) • Don’t want loops, even for short periods of time • Switches 2, 5, 101 send messages immediately, with current age in table • On power-up, switch puts its ports in new pre-forwarding state: sends configuration messages as if designated, but doesn’t forward data 87 (2,1,90) 5(D) 2(B) 78 RP 1(D) (2,1,90) 4(D) (2,1,90) 2(R) 90 3(B) 2(R) 2(R) 66 5 1(R) 3(D) (2,1,5) 2(B) (2,0,2) 2(D) 101 1(D) (2,0,2) 2 2(B) 1(R) (95,0,95) 2(D) (95,0,95) 1(D) 95

  31. Virtual LANs (VLANs) Switched LANs are great, but: • Lack of traffic isolation between LANs • Security, privacy concerns • Inefficient use of switches • One switch per administrative group • Moving users requires physical rewiring VLAN idea: run multiple virtual LANs over a single LAN

  32. Virtual LANs (VLANs) 1 9 15 2 4 8 10 16 … … Computer Science Department Electrical Engineering Department • Traffic isolation: colors = broadcast domains • Easily reconfigurable port assignments • Routing between VLANs: layer 3 routing functionality

  33. The split between L2 and L3 • L2 is point-to-point, L3 has source and destination • Then Ethernet came, world got confused and thought Ethernet was a competitor to L3 • Perlman: “Should have been called ‘Etherlink.’” • Why can’t Ethernet replace IP? • Flat addresses • No hop count (loops would be a disaster) • No fragmentation, reassembly

  34. In each and every host Network adapter Physical card Chip in a device Attaches to host’s system bus Implemented both in hardware and software Software: device driver running on CPU Hardware: FPGA, ASIC in network adapter Where is the link layer implemented? Host Application Memory CPU Transport Network Link Network adapter Controller Link PHY Physical

  35. Hardware or software? Host • Software is more flexible, unless you need: • Time-critical functionality • Backoff transmission timing in Ethernet • Highly parallel processing • Dot product computations for CDMA, filtering for the PHY Application Memory CPU Transport Network Link Network adapter Controller Link PHY Physical

  36. Today • L2 addressing and interaction with L3 • Link layer switching • Learning switches • VLANs • Spanning Tree Protocol • The Point-to-Point Protocol (PPP) • Link virtualization: MPLS

  37. Point to Point link layer • One sender, one receiver, one link: easier than broadcast link: • No Medium Access Control • No need for explicit link layer addressing • e.g., dialup link, ISDN line • Popular point-to-point link layer protocols: • PPP (point-to-point protocol) • HDLC: High level data link control • Data link used to be considered “high layer” in the stack!

  38. PPP Design Requirements [RFC 1557] • Packet framing: encapsulation of network-layer datagram in data link frame • Carry network layer data of any network layer protocol (not just IP) at same time • Ability to demultiplex upwards • Bit transparency: must carry any bit pattern in data • Error detection (but no error correction requirements) • Connection liveness: detect and signal a link failure to the network layer • Network layer address negotiation: endpoints can learn and configure the other’s network address

  39. PPP non-requirements • No error correction nor error recovery • No flow control • No in-order data delivery requirement • No need to support multipoint links (e.g., polling) Error recovery, flow control, data re-ordering all relegated to higher layers!

  40. Framing frames: the problem • We have seen how to frame bits • Ethernet: Manchester encoding, synchronization on edges • Result in general: infinite stream of bits • Problem: where does each frame begin and end?

  41. Frame separators • Choose some pattern of bits as a frame separator e.g.: 1111111 Bit stuffing algorithm Modify the outgoing data stream as follows: • At sender: 111111 → 1111110 • At receiver: 1111110 → 111111 1111111 → (end of frame) Seven 1’s Six 1’s Six 1’s Seven 1’s

  42. PPP: Byte Stuffing • At the sender: • Stuff control escape byte <01111101> before each flag byte occurring in the data • Stuff control escape byte before each control escape byte occurring in the data • At the receiver: • Escape, flag byte  discard escape byte, keep flag byte, continue data reception • Single flag byte: interpret as the flag byte • Two escape bytes  discard one

  43. PPP Byte Stuffing: Example Data in: 01111101 (escape) b4 01111110 (flag) b2 b1 Data out: b1 b2 01111110 (flag) b4 01111101 (escape) PPP Sender PPP Receiver 01111101 01111101 b4 01111110 011111101 b2 b1 stuffed escape stuffed escape

  44. Today • L2 addressing and interaction with L3 • Link layer switching • Learning switches • VLANs • Spanning Tree Protocol • The Point-to-Point Protocol (PPP) • Link virtualization: MPLS

  45. Internetwork layer (IP): • Addressing: internetwork appears as single, uniform entity, despite underlying local network heterogeneity • Network of networks The Internet: virtualizing networks Gateway: • Embed IP packets in local packet format, or extract them • Route (at internetwork level) to next gateway, or end host gateway ARPAnet ALOHAnet

  46. Cerf & Kahn’s Internetwork architecture What is virtualized? • Two layers of addressing: IP and local network • IP makes everything homogeneous • Underlying local network technology • Cable TV links • Satellite links • Telephone modem links • Ethernet, WiFi • MPLS networks • “Invisible” at IP layer: all like a link to IP

  47. Multiprotocol label switching (MPLS) • A large network, viewed as a link, i.e., a type of link layer • Initial goal: Speed up IP forwarding by using fixed length label (instead of IP address) to do forwarding • Borrowing ideas from Virtual Circuit (VC) approach • But: IP datagram still keeps IP address • Insert MPLS header between L2 and L3 headers: PPP or Ethernet header IP header remainder of link-layer frame MPLS header label Exp TTL S 5 1 3 20

  48. MPLS-capable routers • Label-switched routers • Forward packets to outgoing interface based only on label value (don’t inspect IP address) • MPLS forwarding table ≠ IP forwarding table • Signaling protocol needed to set up forwarding • Forwarding possible along paths that IP alone would not allow (e.g., source-specific routing) • Use MPLS for traffic engineering • Must co-exist with IP-only routers

  49. in out out label label dest interface 10 6 A 1 12 9 D 0 in out out label label dest interface in out out label label dest interface 8 6 A 0 6 − A 0 Example: MPLS forwarding tables in out out label label dest interface − 10 A 0 − 12 D 0 − 8 A 1 R6 0 0 D 1 1 MPLS-enabled router R3 R4 R5 0 0 A R2 R1 Standard IP router

  50. Summary: Link layer functionality • Addressing hosts on shared links • The ARP protocol • Error detection and correction (Lecture 2) • Flow control • Framing bits and packets • Reliable delivery point-to-point delivery (next time) • Sharing: Medium access control (Lecture 3)

More Related